Of the roughly half a million Veterans with Alzheimer's disease (AD), the vast majority will progress through the advanced stage of the disease. It is important to ascertain the most safe and effective treatments for individuals with advanced AD. Cholinesterase inhibitor (CI) medications are the mainstay of treatment for mild and moderate AD, and they are often continued in advanced disease. Anecdotal reports indicate that patients may worsen when CI medications are discontinued, but there is little rigorous evidence about the effects of discontinuation. VHA Pharmacy Benefits Management Service Guidelines recommend discontinuing CI under certain conditions, but acknowledge that the issue has not been well studied and that the recommendations are not evidence-based. Because AD follows a variable course, the only way objectively to determine the risks and benefits of CI medications in advanced AD is through a controlled study. We propose to conduct a placebo-controlled, double-blinded discontinuation trial of CI medications in advanced AD. We will recruit 200 Veterans with advanced dementia who have been taking a stable dose of donepezil or galantamine (the main CI medications in the VA formulary), and randomize them. Half the group will carry out a blinded Real Taper: reducing the prior stable dose of CI to one-half for three weeks, then switching to placebo for three weeks. The other half will carry out a blinded Sham Taper: remaining on the prior stable dose for six weeks. We will compare the two arms in (1) ability to complete the 6- weeks successfully, and (2) changes from baseline to week 6 in cognition, functioning, behavior. Analyses will apply on a non-inferiority study design. Specific Aim 1: To ascertain if successful CI discontinuation rates are non-inferior in Real Taper compared to Sham Taper. Hypothesis 1: The rate of successful CI discontinuation in Real Taper will be non-inferior to the rate of successful CI discontinuation in Sham Taper. Specific Aim 2: To ascertain if changes in cognition, functioning, behavior, and caregiver burden over 6 weeks are different in Real Taper compared to Sham Taper. Hypothesis 2: Changes in cognition, functioning, behavior, and caregiver burden in Real Taper will not differ from changes in Sham Taper. Specific Aim 3: To identify patient-level and disease-level factors (such as behavioral symptoms and duration of treatment and disease) that predict unsuccessful discontinuation, adverse events, and worsening. Given the shortage of previous research about CI discontinuation, this aim is not hypothesis-driven. This research will clarify the risks and benefits of CI medications in advanced AD. It will inform guidelines about when to continue or discontinue CI medications, and can thus impact the medical care of tens of thousands of Veterans. More specific information about the effects of treatments for AD will reduce adverse events and maximize benefits. This research will also generate hypotheses about types of individuals for whom discontinuation has differential effects.